This paper reviews the contribution of human PET and SPECT neuroreceptor occupancy studies to the understanding of drug action in psychiatric illness, and how they can aid the development of new drugs. All effective antipsychotics show significant D(2) receptor occupancy. However, at least for atypical antipsychotics, there is no clear relationship between occupancy and clinical response. The mechanisms underlying antipsychotic efficacy, and the minimal effective D(2) occupancy, remain to be elucidated, particularly for drugs with modest or transient occupancy. The low liability of some atypical antipsychotics for extrapyramidal side effects does not appear to be explained by their 5-HT(2A) antagonism, and the muscarinic receptor occupancy of some drugs may be partly explanatory. Previous reports of apparent 'limbic selectivity' of atypical antipsychotics may be in error, and may be due to technical differences in radiotracers. For SSRIs, high occupancies at the serotonin transporter (SERT) are achieved at therapeutic doses, although the minimum SERT occupancy required for therapeutic response remains undefined. Previous attempts to augment the antidepressant effect of SSRIs by pindolol have generally used daily doses which result in inadequate 5-HT(1A) receptor occupancy. For benzodiazepines, clinical doses would appear to leave a wide margin of unoccupied receptors. For methylphenidate and cocaine, typical doses occupy more than 50% of dopamine transporters, and their profiles are extremely similar. In therapeutic drug development, these techniques may be used to assess receptor occupancy profiles, likely drug dosages and dosing intervals which cannot be reliably assessed in humans by other methods.